Communications of the ACM
CLA Professors Stride Forward in Robotic Technology Research
Statistics and computer science professor Song-Chun Zhu creates robots that learn by collecting data from their surroundings.
Credit: Jesse Wang/Daily Bruin
University of California, Los Angeles (UCLA) computer scientists say recent artificial intelligence (AI) innovations are pushing robotic technology forward.
UCLA professor Song-Chun Zhu produces robots that collect images and other environmental data from digital cameras to learn about their surroundings and feed it into statistical models so they can correctly recognize objects they encounter. The method uses digital image processing and natural-language processing to parse images into basic components to draw conclusions about the object's characteristics. Zhu says it enables AI systems to employ a small set of rules to understand a diverse array of situations, similar to the human learning process.
Armed with such models, Zhu's robots can execute high-level tasks and understand complex notions. However, Zhu says more sophisticated AI intelligence is limited by the accuracy of those models, and a computational model of how people think is needed to get machines to conduct human-like tasks.
Most AI scientists are skeptical that anything approaching such a level of robotic intelligence will emerge in the foreseeable future. Still, Zhu and others think current AI systems may prosper when used for specific jobs that demand little advanced reasoning.
From Daily Bruin
View Full Article
Abstracts Copyright © 2016 Information Inc., Bethesda, Maryland, USA
No entries found